Zhou Yahui: 'One‑person companies' are booming; enterprise‑grade Agents are the real gold mine

Thesis: Agents, not chatbots, will remake company structure

Zhou Yahui, chairman and CEO of Tiangong AI (天工AI), told TMTPost that the biggest commercial opportunity in AI is not consumer chatbots but enterprise‑grade productivity Agents that can evolve into an operating system for “one‑person companies.” Kunlun Wanwei (昆仑万维) has just unveiled three flagship models—Matrix‑Game 3.0, SkyReels V4 and Mureka V9—and announced a 2026 “3+1” AGI strategy to support three AI‑native platforms plus a super‑agent aimed at giving individual creators company‑level production power. It has been reported that the company claims these models sit in the global first‑tier; such self‑assessments should be read with caution, but the strategic thrust is clear.

A practical definition of AGI and a market pivot to SaaS

Zhou pushes a pragmatic definition of AGI: the system capability to automate digital and physical workflows, not the science‑fiction general intelligence many imagine. If AGI is judged by the ability to handle finance, sales and growth workflows, then the modern firm’s smallest unit could shrink from teams to individuals. Who needs a chatbot when an Agent can run accounting, marketing and code? He argues this will trigger a deep restructuring of enterprise software (SaaS) — legacy SaaS will be “eaten” by AI, and the next wave of Agents, particularly enterprise‑grade ones, will capture the real value.

Costs, competition and China’s edge — but not without risks

Zhou warned that model economics are changing: token pricing will be scene‑specific (code and finance tokens pricier; general models cheaper), and compute scarcity overseas already makes token subsidies impractical in some markets. It has been reported that Kunlun Wanwei’s overseas revenue accounts for over 90% of sales and that the company plans at least RMB 100 million per month in R&D investment — a sign of the cash‑intensive race ahead. Geopolitics matters: export controls, chip and cloud access constraints, and Western competition shape where Chinese firms can play. Zhou’s answer to maintaining an edge is simple: endurance in hard R&D. But he also cautions about structural problems inside Chinese tech organizations — decision-makers often lack deep technical literacy — a gap that could blunt original innovation even as companies scale global ambitions.